Active conflict
Threat scoring, air-defence alerts, GDACS escalations, and the OSINT signals that drive them.
A working open-source picture of an active conflict needs three things: a stable view of the battlefield geography, a stream of fresh signals telling you what just happened, and a way to score the noise so the operator looks at the right thing first. SENTINEL ties these together with country threat scores, region-level alert overlays, and a classifier that reads the incoming signal stream against a known event grammar.
Country threat scores are computed from a base level plus modifiers for active conflict-zone designation, fresh strikes, dynamic signals from X / Telegram / wire services, and any active airspace closure. Each score component is visible on the country panel, so the operator can see exactly why a country is amber rather than green and which input is doing the work. The minimum score floor of ten ensures even quiet countries carry a baseline reading rather than disappearing.
Air-defence alerts are the highest-frequency live overlay. Ukrainian oblast alerts render as red-striped polygons that fade with zoom; Israeli Tzeva Adom city alerts render as smaller circles tuned for the 3-kilometre warning radius. Both feeds carry alert type, start time, and source, and dropping the chip pin opens a card with the latest co-occurring sub-region alerts. Russian and Iranian air-defence activations, when public, surface alongside.
The Global Disaster Alert and Coordination System (GDACS) renders escalation polygons for non-conflict events that nonetheless reshape a security picture: hurricanes, floods, earthquakes, volcanic eruptions, wildfire complexes. SENTINEL keeps these in the same scoring pipeline so a country dealing with a major disaster shows the strain in its dynamic score even when no military activity is in play.
GPS spoofing and jamming sit at the seam between conflict and electronic warfare. Spoofing clusters render as hex-binned overlays with intensity. Routine background interference around the usual hot zones is filtered; a genuine step-change at a specific flashpoint surfaces as a live event. Receiver-handoff false positives in mid-Atlantic / mid-Pacific gaps are aggressively filtered so they do not pollute the rail.
For the breaking-news case, the noteworthy events panel lists every X and Telegram post the classifier has flagged as cinematic-worthy in the last hour. Each item carries the source post text, attached image or video, country tag, and a fly-to button that rotates the globe to the event location. This is the fast path from "something happened on Twitter" to "I am looking at the location on a map."
What to look for on the globe
- Threat-score breakdown
The country panel shows base score, dampening, signal delta, and airspace closure modifier so the math reconciles.
- Tzeva Adom + Ukraine
City-level Israeli alerts and oblast-level Ukrainian alerts share the same renderer but tuned for their respective scale.
- GPS spoofing
Hex bins with intensity. Mid-ocean false positives filtered; real step-changes near flashpoints surface.
- Cinematic + noteworthy
The noteworthy panel and cinematic loop both consume the same classifier output, just at different cadences.